A Novel 2D-to-3D Conversion Method Based On Motion Estimation
With the development of 3D display technology and the success of 3D films, more and more attention and desire is drawn to 3D video industry. The conversion of existing 2D videos to 3D videos plays a significant role in the 3D content production. This work presents a novel method that is able to convert 2D videos to 3D videos automatically based on the motion information. The proposed method carries out pyramid segmentation algorithm with the key frames and estimates their optical flow. A depth map is then assigned utilizing the motion parallax cues. Next, the depth map is modified by filtering and morphology operators to diminish existing block artifacts and generate more visually comfortable depth map. Finally, the depth map is ready to be used to produce a 3D image of the key frames using a depth image of the key frames asing a depth image based rendering method1.
2d to3d depth map optical flow morphology
Yudong Guan Bowen Huang Mingjuan Liu Lili Sun Chunli Ti
School of Electronics and Information Engineering Harbin Institute of Technology Harbin, China Daqing Oil Field Real Estate Development Co., Ltd Daqing Oil Field Power Supply Company of Electric Power Group Daqing, China
国际会议
三峡
英文
1319-1322
2012-05-18（万方平台首次上网日期，不代表论文的发表时间）